Background Preoperative staging of cervical lymph nodes is important to determine the extent of neck dissection in patients with oral squamous cell carcinoma (OSCC). Purpose To evaluate whether a preoperative discrimination of benign and malignant cervical lymph nodes with diffusion-weighted imaging (DWI) (3T) is feasible for clinical application. Material and Methods Forty-five patients with histological proven OSCC underwent preoperative 3T-MRI. DWI (b=0, 500, and 1000s/mm(2)) was added to the standard magnetic resonance imaging (MRI) protocol. Mean apparent diffusion coefficients (ADC(mean)) were measured for lymph nodes with 3mm or more in short axis by two independent readers. Finally, these results were matched with histology. Results Mean ADC was significantly higher for malignant than for benign nodes (1.1430.188 * 10(-3) mm(2)/s vs. 0.987 +/- 0.215 * 10(-3) mm(2)/s). Using an ADC value of 0.994 * 10(-3) mm(2)/s as threshold results in a sensitivity of 80%, specificity of 65%, positive predictive value of 31%, and negative predictive value of 93%. Conclusion Due to a limited sensitivity and specificity DWI alone is not suitable to reliably discriminate benign from malignant cervical lymph nodes in daily clinical routine. Hence, the preoperative determination of the extent of neck dissection on the basis of ADC measurements is not meaningful.